Expand `reportUninitializedInstanceVariable` to protocol variable implementation detection

```python
class MyProtocol(Protocol):
a: int
def f(self) -> None:
self.a = 0

@final
class MyClass(MyProtocol): # want an error here
pass

class MyClass2:
def f(self) -> None:
self.a = 0 # error: reportUninitializedInstanceVariable

x: MyProtocol = MyClass2() # want an error here
```
Currently, `reportUninitializedInstanceVariable` setting is not applied to protocol variable implemetation detection for both implicit and explicit protocol implementation.

For implicit implementation, `reportUninitializedInstanceVariable` error appears in the method but this is not an error indicating the class does not implement the protocol.

It's really helpful if `reportUninitializedInstanceVariable` prevents protocol variable implementation outside of `__init__`.

Start by reproducing the implicit and explicit protocol examples with reportUninitializedInstanceVariable enabled, then trace the protocol variable implementation detection entry point and its handling of that setting. Done means both examples produce the intended protocol implementation error rather than only the method-local diagnostic.

Written by the indexing model from the issue text.